Do you have a pet project?

I do, and I don't like the fact that I've abandoned most of them before finishing. My current pet project helps me collect interesting materials and organize them by relevance. It's a mix of a Telegram bot where I can post interesting links, a web crawler that gathers links automatically, and a simple ML model that acts as the recommendation engine. This project has already saved me a lot of time, and I don't like the idea of leaving it unfinished.

How do you collect your ideas?

In my case, I needed a commitment. A simple todo.md file committed to my repo worked pretty well. It was a good start, but it had its limits and, honestly, wasn't all that creative. So I switched to Jupyter notebooks. First, they let me do quick ad-hoc calculations; second, they support Markdown. I put together a list of milestones like this:

### Oh-my-reading-list next steps
- Switch post embeddings model
- Implement data fetcher for MDN blog
- Implement data fetcher for developer.chrome.com blog
- Implement data fetcher for web.dev
- Implement data fetcher for Evil Martians blog
- Update rec model to never ever suggest vibecoding articles
- Add quick summary of the post

When I have a spare hour, what should I pick? I stared at the list, thinking deeply, and after an hour finally selected the best item...

Try two: I ordered the list once and went step-by-step. I picked the first task, worked on it for an hour, but didn't finish. A few days later, I'd forgotten all the context and had to start over. And again, and again...

Clearly, I needed to reorder my tasks. I decided to assign two extra numbers to each task:

1. "coolness" - how cool my project becomes after implementing this task (in sunglasses)
2. "complexity" - how complex this task is (in penguins)

Unlike real-life projects, it's totally fine to convert penguins to hours here, since I'm not interacting with other people :) So, one hour equals 5 penguins. What should I pick next? The answer is simple: estimate your spare hours for the next week and solve the knapsack problem.

const readingList = [
    { title: "Implement data fetcher for developer.chrome.com blog", coolness:
5, complexity: 1 },
    { title: "Implement data fetcher for web.dev", coolness: 5, complexity: 1 },
    { title: "Implement data fetcher for MDN blog", coolness: 5, complexity: 1 },
    { title: "Implement data fetcher for Evil Martians blog", coolness: 5, complexity: 1 },
    { title: "Update rec model to never ever suggest vibecoding articles", coolness: 10, complexity: 25 },
    { title: "Add implement quick summary for the articles", coolness: 3, complexity: 10 },
    { title: "Switch post embeddings model", coolness: 10, complexity: 25 }
];

const spareHours = 3;

const knapsack = (items, capacity) => {
    const n = items.length;
    const dp = Array.from({ length: n + 1 }, () => Array(capacity + 1).fill(0));

    for (let i = 1; i <= n; i++) {
        for (let w = 0; w <= capacity; w++) {
            if (items[i - 1].complexity <= w) {
                dp[i][w] = Math.max(dp[i - 1][w], dp[i - 1][w - items[i - 1].complexity] + items[i - 1].coolness);
            } else {
                dp[i][w] = dp[i - 1][w];
            }
        }
    }

    const result = [];
    let w = capacity;
    for (let i = n; i > 0 && w > 0; i--) {
        if (dp[i][w] !== dp[i - 1][w]) {
            result.push(items[i - 1]);
            w -= items[i - 1].complexity;
        }
    }
    return result.reverse();
};

const selectedTasks = knapsack(readingList, spareHours * 5);
console.log("Selected tasks for the week:", selectedTasks);

The output will be:

Selected tasks for the week:
[
    {title: 'Implement data fetcher for developer.chrome.com blog', coolness: 5, complexity: 1},
    {title: 'Implement data fetcher for web.dev', coolness: 5, complexity: 1},
    {title: 'Implement data fetcher for MDN blog', coolness: 5, complexity: 1},
    {title: 'Implement data fetcher for Evil Martians blog', coolness: 5, complexity: 1},
    {title: 'Add implement quick summary for the articles', coolness: 3, complexity: 10}
]

Alternatively, I can sort the tasks by efficiency and just pick the most efficient ones:

const sortedTasks = readingList.sort((a, b) => (b.coolness / b.complexity) - (a.coolness / a.complexity));
console.log("Selected tasks one by one:", sortedTasks);

I can also visualize the tasks in 2D space. This helps me to find the most effective tasks and understand the overall design of my project. The steeper the curve the better!

Conclusion

Sometimes adding a bit of data in unexpected places turns out to be useful, however be aware. I spent an hour writing this post. Guess how many tasks for my pet project I've finished? ;)